Chapter 8: Concurrency

Concurrency is a question about the completion date, so it starts with what moves that date. The critical path is the longest chain of dependent activities through the job, and there can be more than one. Its length fixes the finish. A delay to a critical activity will push completion back unless the contractor accelerates or re-sequences the work. A delay to any other activity first uses up its float: the time it can slip before it holds up anything else. Chapter 2 explains both ideas.

Only critical delay counts, and the reason is ordinary causation. An extension of time answers one question: how much later will the job finish because of the employer's event? An event that delays only non-critical work does not make the job finish later, so it earns no extension. The prevention principle works the same way. It stops an employer holding the contractor to a date the employer has prevented it from meeting. Hamblen J pointed out that this necessarily means prevention in fact, not in theory (Adyard [264]). An employer's act that causes no delay to completion gives no time.

So a delay off the critical path cannot be a concurrent cause. English cases ask whether each event in fact delayed completion (Adyard [279]; Thomas Barnes [140]). Singapore's Appellate Division put it shortly in ICOP: it is meaningless to speak of concurrent causes of a delay that is not on the critical path ([100]). The SCL Protocol adds that, generally, the completion date moves only once float is used up (Core Principle 8).

Judges define concurrent delay using a 2002 article by John Marrin QC: a period of overrun caused by two or more effective causes of delay which are of approximately equal causative potency. The Court of Appeal adopted it in North Midland in 2018, as first-instance judges had before. It did so in a passage it described as only tangentially relevant to the appeal ([16]). In practice the definition means that each cause must in fact have delayed completion over the same period, and neither must swamp the other.

Concurrency needs its own rule because the ordinary test of causation gives no answer. The but-for test asks whether the delay would have happened without the event. Take away the employer's event and the job is still late, because of the contractor's. Take away the contractor's event and the job is still late, because of the employer's. Neither cause passes the test, so on a strict reading neither caused the delay. Yet the job is late, and someone has to bear the period. The courts have had to choose a rule, and they have chosen different rules for time and for money.

The SCL Protocol, the Society of Construction Law's guidance on delay, uses the word in two senses. The first, true concurrency, is two events at the same time whose effects are felt at the same time, and it will be a rare occurrence. The Protocol's example is the first day on site, when the employer fails to give access and the contractor has nothing mobilised (10.3). More often the word covers events that arise at different times but whose effects are felt at the same time (10.4). In both senses each event must be an effective cause of delay to completion, not merely incidental to it (10.5).

Delays at different times are not concurrent at all. Suppose the employer's event holds up the critical path in weeks 10 to 12 and the contractor's in weeks 20 to 22. Each side bears its own. The contractor gets 3 weeks of extension for the first, and its costs if that event carries money. It pays 3 weeks of liquidated damages for the second.

Time and money get different answers because they answer different questions. The extension of time asks whether the employer may hold the contractor to the original date, and charge liquidated damages, for delay that the employer's own event caused. The money claim asks whether the employer must pay for the contractor's loss. That needs proof that the employer's event caused the loss. In a concurrent period the employer cannot charge liquidated damages, and it need not pay the contractor's costs.

Take time first. An extension clause exists mainly to relieve the contractor from liquidated damages for delays which were not his responsibility (North Midland [44]). In a concurrent period the employer's event did delay completion. Without an extension, the employer would collect liquidated damages for a delay its own event helped to cause. That the contractor would have been late anyway does not change this. The contractor must have a reasonable time within which to complete, and the employer has taken part of that time away (De Beers v Atos [177]).

The courts reached the same answer through the words of the JCT extension clause. In Walter Lilly (TCC, 2012) Akenhead J's main reason was that the clause extends time for the whole delay the Relevant Event caused. Nothing in it cuts the extension down because another cause was running too. He added that many Relevant Events would otherwise be acts of prevention. The certifier (the architect under JCT, the engineer under FIDIC, the project manager under NEC) must fix a "fair and reasonable" extension. That does not mean splitting the period: the test is primarily a causation one, and the contractor gets a full extension ([370]). Chapter 3 covers the prevention principle and the machinery for extending time.

The rule on time goes back to Malmaison (1999). Dyson J gave an example the parties had agreed. No work is possible on site for a week, because of exceptionally bad weather (a Relevant Event) and because the contractor is short of labour (not one). If that week will delay completion by a week, the architect, if he considers it fair and reasonable, must grant a week. He cannot refuse on the grounds that the delay would have occurred in any event. Dyson J's words are quoted as Hamblen J set them out in Adyard [276].

Now money. Unless the contract says otherwise, loss and expense, like damages, is recoverable only for loss the employer's event caused. In a concurrent period the contractor's own delay would have kept it on site for the same weeks at the same cost. The employer's event made no difference to that loss, so the but-for test fails. In De Beers (TCC, 2010) the judge said the contractor cannot recover where he would have suffered exactly the same loss from causes for which he is responsible ([178]). The contractor's time-related costs for the concurrent period stay with it.

The two halves fit together. Without a rule, each side would have a money claim against the other for the same weeks: the employer for liquidated damages, the contractor for its prolongation costs. As Marrin wrote in 2013, in the absence of apportionment, both claims cannot succeed. He argued that the English approach, time as in Malmaison and ordinary proof of causation for money, deals with that problem (p.17). The employer loses its liquidated damages for the period, and the contractor bears its own costs.

Time without money is familiar anyway. Bad weather is the SCL Protocol's example. It is at the employer's risk for time, but carries no entitlement to compensation for prolongation (12.2). Chapter 3 sorts the standard events into time, money or both.

The authorities state both halves in a line. De Beers concerned a failed IT system, not a building, but Edwards-Stuart J put the rule in terms of construction and engineering cases:

He applied it: the supplier would have had an extension for one bundle of work, but no prolongation costs for the same delay ([179]). In Thomas Barnes (TCC, 2022) counsel agreed in closing submissions that the law is as Keating (11th edition, 9-105) states it. Depending on the precise wording of the contract, a contractor is probably entitled to an extension if the event relied on was an effective cause of delay. It recovers loss and expense only where it satisfies the but-for test ([118]). The SCL Protocol recommends the same answers (Core Principles 10 and 14). These are first-instance decisions, and how firm the rule is is a separate question.

Thomas Barnes shows the rule in figures. The case concerned a bus station in Blackburn whose construction ran badly over. The Council was the employer. The steelwork for the hub, which was the Council's risk, deflected and needed remedial work, and the contractor was late with the roof coverings. HHJ Stephen Davies found that both items were on the critical path and both were causing delay over the same period ([140]). The further extension came to 119 days: 133 days for the steel frame, less 2 days mitigated and less 12 days for the late start of the remedial works ([148]). Prolongation money was allowed for only 27 of those 119 days ([157]).

Two limits on the case matter. The law was agreed rather than argued, and an amended clause reinforced the refusal of money ([147]; see writing your own rule). So the case applies the rule rather than tests it. And in the end the contractor recovered nothing. The cost of completing the job after the Council terminated more than wiped out anything due to it ([14]).

The model project used throughout this guide shows the same rule. It is invented and illustrative: thirteen activities, planned finish week 34, contract completion week 36. A variation adds 4 weeks to the M&E installation and, at the same time, the cladding starts 8 weeks late. Each alone would push the finish to week 38. On the English default rule the contractor gets 2 weeks of extension, pays no liquidated damages, and gets no money for the concurrent weeks.

Three further points on money come from the SCL Protocol. First, on its approach the contractor is in most cases paid only for any period by which the employer's delay exceeds its own (14.3). Separating the two usually needs as-planned and as-built programmes with their critical paths, and records of which extra work was at whose risk (14.4; chapter 14). Second, the Protocol values the sum for the period in which the employer's event was felt, not the extended period at the end (22.3; chapter 13). Third, the Protocol warns employers about a variation instructed after the completion date while the contractor is in culpable delay. If the contractor then accelerates at its own cost and the variation becomes the effective cause of the late finish, the employer may lose its liquidated damages (10.13; chapter 10).

The rule on time and money applies only if both delays in fact drove completion over the same period. This guide calls that requirement the gateway, and it comes from ordinary causation again. An extension is given for the delay an employer's event causes to completion. Suppose the contractor's own delay has already fixed a later finish, and the employer's event then arrives and changes nothing. That event has caused no delay to completion. An event that causes nothing gives nothing, just as a breach that causes no loss gives no damages. So there is no extension and no concurrency. Giving time there would release the contractor from liquidated damages for a delay that was its own. Hamblen J said the act relied on must cause some actual delay (Adyard [282]).

The classic statement is in Royal Brompton (No 7), a TCC decision of HHJ Richard Seymour QC in 2000. A new heart and chest hospital in Chelsea was due for completion on 23 July 1989. The architects extended time because the contractor had been given part of the site late, and in the end time ran to 22 May 1990. The hospital later sued its architects. It said no extension should have been given for the late release at all: the contractor was already so far behind that the release could not have caused critical delay. The judge said concurrency does not describe a case where work is already delayed by the contractor's labour shortage and a Relevant Event then occurs which in fact made no difference. Real concurrency is where the works are on programme and two things happen, either of which alone would have delayed them ([31]).

Hamblen J turned the passage into a test in Adyard, a 2011 Commercial Court case about building ships. There is concurrency only if both events in fact cause delay to the progress of the works and their delaying effect is felt at the same time ([279]). The shipyard lost. It had to prove causation in fact for the buyer's variations, and the project was already in irretrievable critical delay before they arrived ([292], [294]). Delay that existed only on paper was not enough.

In Saga Cruises (2016) a deputy judge of the Commercial Court called Hamblen J's treatment perhaps the most useful synthesis and applied it to a cruise-ship refit on bespoke terms. Unless a concurrency actually affects the completion date as then scheduled, the contractor cannot claim the benefit of it ([251]). She held that owners' items arriving while the yard was already late could not be relied on as concurrent delay ([301], [303], [313]). All three cases are first-instance decisions, and each confines the rule rather than applies it.

The SCL Protocol works the same point through with dates. A contractor's event delays completion from 21 January to 25 February. A later variation would, on its own, have delayed completion from 1 to 14 February. On one view both are effective causes for those two weeks, because each would have delayed completion without the other (10.8). On the other view the variation does not make the works finish later than they otherwise would, because they were already going to finish later still (10.9).

The Protocol recommends the second view as the consistent position of recent lower-court English decisions. It warns that the recommendation would need reconsidering if an appeal court took a different approach (10.9, 10.10). The second view is sometimes called "first in time", but the Protocol does not use that label. What matters is which delay had already fixed the later completion date, not which event came first.

Two further points go with the example. An employer's delay does not exonerate the contractor for its earlier delays, and any extension is simply added to the contract completion date (10.14). That is the "net" method the court adopted in Chestermount (Walter Lilly [364]), which chapter 3 explains under the net method. And delay analysis is rarely precise down to the day, so the Protocol says the margin for imprecision should be taken into account in deciding whether delays were concurrent (10.11). An expert can use that margin to find concurrency or to deny it.

Float can decide the gateway, but it does not always. In Thomas Barnes the Council's delay expert said the steelwork had enough float to stay non-critical until about 9 December 2014; before then the roof alone was driving completion ([141]). The judge accepted that evidence from a theoretical delay analysis viewpoint, but not as an answer on causation. Both items were in fact holding up the hub finishes over the same period ([140], [143]). Nor could the contractor say that all the delay between October 2014 and January 2015 was the steelwork's, just because the problem ran between those dates ([144]). Both were driving, so the rule applied.

Pacing raises the same question from the contractor's side. A contractor that knows the employer's delay has already pushed completion back may slow its own non-critical work rather than hurry up and wait. In Tata v DBS (TCC, 2024), an IT project, Constable J described pacing as a non-critical activity prolonged deliberately in the face of delays to a critical one. Whether it happened is a question of fact ([180]). The pacing case failed. No witness or document showed a decision to slow down, and the contractor's own expert accepted that his examples were not really pacing ([182]). The SCL Protocol recommends that a contractor who intends to pace tell the employer and the certifier, with its reasons (15.2). I would do that at the time, in writing.

The rule on time began as common ground. In Malmaison the concurrency passage opens with the words it is agreed that. Akenhead J accepted that Dyson J could be said to have repeated the common ground rather than decided the point. He added that Dyson J seemed to endorse it (Walter Lilly [367]).

The Court of Appeal has said how far that goes. In North Midland Coulson LJ noted that the point in Malmaison was conceded. A contractor's entitlement to an extension for concurrent delay under the unamended JCT forms is not entirely free from doubt, and there is no Court of Appeal authority on it ([17]). He set out the contrary view in Keating (10th edition): where the causes are concurrent, the prevention principle would not be triggered because the delay would have occurred anyway. He noted that two first-instance decisions are cited for that view, Adyard and Jerram Falkus ([17]). Counsel for the employer raised a wider argument along the same lines. The court declined to decide it: the point did not arise, and it would be unwise to decide it without full argument ([50]).

What holds the rule up is two first-instance decisions, De Beers [177] and Walter Lilly [370]. Hamblen J's reading of Malmaison in Adyard [277] supports them, but it was obiter, in a case the yard lost. Thomas Barnes [118] adds an agreed statement of law. A High Court decision in the TCC or the Commercial Court is persuasive and often followed, but it does not bind other High Court judges or the Court of Appeal. The Court of Appeal refused permission to appeal in Royal Brompton, calling the judgment exemplary. Refusing permission does not make [31] Court of Appeal authority. As at September 2026, the date at which this guide states the law, no appeal court in England and Wales has decided the point.

Other courts have not settled it either. Singapore's Appellate Division was asked to choose between the Malmaison line and what the appellant called the "first in time" approach of Saga. It found the debate irrelevant, because neither delay was critical (ICOP [92], [100]).

The gateway has critics too. Lord Carloway, dissenting in City Inn, thought the Royal Brompton view appears to be in error ([110]). The Lord Ordinary in the same case had treated overlap as true concurrency regardless of which cause started first ([50]). These are Scottish views on the JCT clause, and English courts have not followed them. No English appeal court has decided the point.

The prevention principle, argued on its own, gives the contractor less. In Jerram Falkus (TCC, 2011) the court held that the principle will not apply where the contractor's own concurrent delays mean the earlier date would not have been met anyway ([52]). Where the extension clause covers the employer's event, the contractor still gets its full extension under the clause (Walter Lilly [370]). The principle is then not engaged at all (North Midland [31]). The Court of Appeal left the relationship between prevention and concurrency open ([50]).

An employer with enough at stake therefore has room to argue. The Court of Appeal noted that the possibility of a full extension for concurrent delay has led employers to write their own answer into the contract (North Midland [18]).

The default rule is the law's answer where the contract says nothing, so the parties can agree a different one. Who bears a period of concurrent delay is a question of risk, and allocating risk is what contracts do. A clear concurrency clause will therefore be enforced, and the Court of Appeal has said so. In North Midland the parties had amended the JCT Design and Build 2005 form. The new clause read:

Coulson LJ held that the clause allocated the risk of concurrent delay to the contractor. That was an allocation of risk which the parties were entitled to agree ([22], [39]). It did no more than reverse the result in Malmaison and Walter Lilly for that contract ([35]).

The contractor argued that the prevention principle overrode the clause. Coulson LJ rejected that for five reasons ([29]). The principle is not an overriding rule of public or legal policy ([30]). It was not engaged, because the contract gave an extension for the employer's acts of prevention ([31]). None of the old prevention cases mentions concurrent delay ([32]). And, in what he called perhaps the most important reason of all, parties can contract out of some or all of the principle's effects ([36]). Liquidated damages then run for the concurrent period. The court will not imply a term that contradicts the express words to take them away ([45]). Either result may be harsh on the other party, he said, but neither is uncommercial ([47]).

TTSJV v BapCo shows one large contract doing the same. It was a 2026 interim decision of the TCC on an engineering, procurement and construction contract governed by English law. Its clause 8.3(d) excluded any extension where there were concurrent delays and at least one of them would not itself give an entitlement ([46]). The judge thought, without hearing argument on it, that a further clause required the parties to give effect to the employer's rejection of the claim while it was challenged ([49]). The decision was on an urgent injunction application, so it illustrates the drafting and decides nothing about concurrency.

A general "no extension for contractor default" clause is not a concurrency clause. In Thomas Barnes clause 2.29A said the contractor would not be entitled to any extension for a circumstance arising from its own error, omission, negligence or default ([76]). The judge held that it did not alter the analysis on time, because the steelwork delay was not such a circumstance ([146]). But it reinforced the converse conclusion: no loss and expense for the concurrent period ([147]).

An apportionment clause helps only if a party pleads it and uses it. In Tata v DBS the contract dealt with a delay partly caused by the contractor's default and partly by a cause at the authority's risk. It said that the parties shall negotiate in good faith with a view to agreeing a fair and reasonable apportionment of responsibility ([50]). The contractor argued, late, that this gave an English court the power to apportion delay that exists under Scots law ([101]). Constable J held that a case on apportionment ought to have been pleaded, so that the witnesses and experts could deal with it. There was also no evidence that the parties had negotiated as the clause required ([104]). Whether such a clause could give a court that power was not decided.

Scotland asks a different question. In England the extension turns on causation: if the employer's event caused the delay, the contractor gets all of it, whatever else was going on. Where neither cause is dominant, Scots law lets the decision-maker weigh the two and share the period between them. On the English view, a share would give the contractor less time than the employer's event in fact cost it. Akenhead J named the two schools in Walter Lilly. Under the English approach the contractor gets a full extension for delay caused by two or more events, provided one is a Relevant Event. Under the Scottish approach it gets a reasonably apportioned part ([366]). Apportionment means splitting the concurrent period, or its cost, between the two causes.

The Scottish case is City Inn, decided by the Inner House of the Court of Session in 2010. A hotel in Bristol was built on the JCT 1980 private form. The contractor claimed 11 weeks of extension for the architect's late instructions, and it had its own troubles with the lifts and the stair balustrades. The Lord Ordinary found that none of the causes was dominant, and he gave 9 weeks. The appeal failed.

Lord Osborne set out the propositions ([42]). A Relevant Event must be shown to have caused delay to completion, and causation is a matter of common sense on the evidence. A sound critical path analysis helps, but its absence is not fatal. If a dominant cause can be identified, effect is given to it and the other causes are left out of account. Where two causes operate and neither is dominant, the claim does not necessarily fail. Instead:

Lord Osborne also said that asking whether there is true concurrency does not assist under the JCT clause ([51]). The decision carries less weight than it seems. Lord Kingarth agreed in one sentence ([98]), and Lord Carloway dissented on apportionment. Take 6 weeks from a Relevant Event and 6 weeks from a labour shortage over the same period. In his view the architect fixes a date six weeks on, and that is not an apportionment exercise ([113]). "Fair and reasonable" goes to fixing the new date once causation is decided ([114]). On that point his dissent matches the English rule. No judge fixed a proportion; the 9 weeks out of 11 was a judgment on the facts.

English courts have refused apportionment twice. Akenhead J held that City Inn, although of persuasive weight, is inapplicable in England (Walter Lilly [370]). In Tata v DBS Constable J held that it is not open to the court generally to apportion responsibility for delay. On the evidence it could not have been done reliably anyway ([229]). So there is no general power to apportion in England. A contract can build its own mechanism, but the one in Tata was never pleaded or used.

Adyard rejected a different strand of City Inn. Hamblen J read Lord Carloway as suggesting that a Relevant Event need not cause actual delay, and held that this does not reflect English law. Under the JCT form and similar contracts the Relevant Event must be at least a concurrent cause of actual delay to the progress of the works ([286]).

Apportionment came into Scots law through money, in John Doyle (Inner House, 2004), a case about a global claim. Where the loss cannot be traced to each cause and no cause is dominant, the court said the loss may be split between the causes. It accepted that the result may be somewhat rough and ready, much like splitting blame for contributory negligence ([17]). It suggested an equal split where the competing cause is not the contractor's responsibility. Where it is, it may be right to deny the contractor any recovery for the period ([16]). That was a decision on the pleadings, so it is guidance, not a rule. Chapter 12 has John Doyle on global claims.

In City Inn the Lord Ordinary relied on John Doyle for the money too. He split the prolongation costs by weighing, as he had for time, how much each cause contributed and how culpable each side was. He gave 9 weeks of them ([60]). The Inner House upheld that approach ([62], [63]). He did not apply the rider about denying recovery, although the competing causes were the contractor's.

Beyond Scotland, assume nothing. FIDIC's 2017 Red Book (first printing) does not choose a rule. Where an employer-risk delay is concurrent with a contractor-risk delay, the extension is assessed under the rules and procedures stated in the Special Provisions. If there are none, it is assessed as appropriate, taking due regard of all relevant circumstances (Sub-Clause 8.5). FIDIC's guidance explains that there is no one standard set of rules in use internationally. The JCT Design and Build Contract 2016 and NEC4 have no express concurrency clause; chapter 16 compares the forms. Singapore's Appellate Division has left the choice between approaches open (ICOP [101]). On a project governed by another law, check that law and the Special Provisions before assuming the English rule travels.

The extension that concurrency gives is worked out under the machinery in chapter 3, and the money it withholds is what chapter 13 prices. Whether there was concurrency at all is a question of proof: chapter 6 and chapter 14.